Nagios Support Forum

One of the big problems we're finding has to do with the performance graphs. Every time we change the service description on a monitor it breaks the performance graphs. The other challenge is merging the performance graph data back. I haven't found any good tools to do that yet. There are two tools I googled for, one python merge tool and another perl tool. Neither of them worked.

Here are a few things that would help solve this problem:
1) Don't tie the rrd graphs into the service description, tie them into the database indexes instead. That way they don't change very often.
2) If a graph breaks for any reason, you lose all the data, which is a bad thing when it's all we have to rely on for outages sometimes.
3) If there is a good utility for merging graphs, I need it to help get some information about an outage we experienced this last week.

Let me know what I can do. Thanks!

That seems like a reasonable list of points. I know there is a new upstream release of the graphing component that we're working on integrating, so it will be interesting to see if it addresses some or all of those.

I know you guys hate this question sometimes, but any ideas when the new release it coming out with this particular fix?

No, but I doubt it would be terribly soon, since that would be a major version bump. Possibly by year end though - depends which one this goes in.

Any word on this one? We're going to a service name change and would really like something for this.

We're forking off the code repositories this week or next in preparation for a new major release line. I don't know much in the way of details yet still, but this will give us a better chance to test these sorts of changes at least.

The challenge with this particular issue is that both the rrdtool and PNP use the service description as the key identifier for that data. When performance data is returned, it puts the rrd file for that service into a folder of the host's name, and then names it based on the service description. PNP also uses the service description as a URL argument to show the generated graph. I'll do some hunting and see if there's any way to preserve the data from the old name and maybe merge it with a new file.

I did some checking on what happens to the rrd file once a service description gets changed. It looks like the old data is still there and is accessible. You can even view the graph of it directly from PNP as well. http://<yourserver>/nagios/pnp/ and then search for that host. You should be able to view a graph for both the old service name, and the new one.

As far as merging tools, you probably already found the same results I did on merging tools. There doesn't appear to be very good solutions for it.

Nagios is built around organizing the checks by host_name->service description, as are most of the tools built for Nagios. I'm not sure we have a perfect solution for you at this time...

mguthrie wrote:I did some checking on what happens to the rrd file once a service description gets changed. It looks like the old data is still there and is accessible. You can even view the graph of it directly from PNP as well. http://<yourserver>/nagios/pnp/ and then search for that host. You should be able to view a graph for both the old service name, and the new one.

As far as merging tools, you probably already found the same results I did on merging tools. There doesn't appear to be very good solutions for it.

Nagios is built around organizing the checks by host_name->service description, as are most of the tools built for Nagios. I'm not sure we have a perfect solution for you at this time...

Ok the nagios/pnp site is pretty good. I can't believe there's no good merging utilities, though. I found one written in python that didn't work. Although, I like the new functionality on the graphs to drill down to the exact hour. Although it would be nice to be able to specify the times in the graphs themselves.

We actually just had another user submit some code that has selectable timeframes to view graphs in. I've got it on my TODO list to take a look at that code and get it turned into a component. However, that component would be for the XI interface only.